Two branches of Tangjiagou rock avalanche were triggered by Lushan earthquake in Sichuan Province,China on April 20th,2013.The rock avalanche has transported about 1500000 m3 of sandstone from the source area.Based on...Two branches of Tangjiagou rock avalanche were triggered by Lushan earthquake in Sichuan Province,China on April 20th,2013.The rock avalanche has transported about 1500000 m3 of sandstone from the source area.Based on discrete element modeling,this study simulates the deformation,failure and movement process of the rock avalanche.Under seismic loading,the mechanism and process of deformation,failure,and runout of the two branches are similar.In detail,the stress concentration occur firstly on the top of the mountain ridge,and accordingly,the tensile deformation appears.With the increase of seismic loading,the strain concentration zone extends in the forward and backward directions along the slipping surface,forming a locking segment.As a result,the slipping surface penetrates and the slide mass begin to slide down with high speed.Finally,the avalanche accumulates in the downstream and forms a small barrier lake.Modeling shows that a number of rocks on the surface exhibit patterns of horizontal throwing and vertical jumping under strong ground shaking.We suggest that the movement of the rock avalanche is a complicated process with multiple stages,including formation of the two branches,high-speed sliding,transformation into debris flows,further movement and collision,accumulation,and the final steady state.Topographic amplification effects are also revealed based on acceleration and velocity of special monitoring points.The horizontal and vertical runout distances of the surface materials are much greater than those of the internal materials.Besides,the sliding duration is also longer than that of the internal rock mass.展开更多
In the field of hyperspectral image(HSI)classification in remote sensing,the combination of spectral and spatial features has gained considerable attention.In addition,the multiscale feature extraction approach is ver...In the field of hyperspectral image(HSI)classification in remote sensing,the combination of spectral and spatial features has gained considerable attention.In addition,the multiscale feature extraction approach is very effective at improving the classification accuracy for HSIs,capable of capturing a large amount of intrinsic information.However,some existing methods for extracting spectral and spatial features can only generate low-level features and consider limited scales,leading to low classification results,and dense-connection based methods enhance the feature propagation at the cost of high model complexity.This paper presents a two-branch multiscale spectral-spatial feature extraction network(TBMSSN)for HSI classification.We design the mul-tiscale spectral feature extraction(MSEFE)and multiscale spatial feature extraction(MSAFE)modules to improve the feature representation,and a spatial attention mechanism is applied in the MSAFE module to reduce redundant information and enhance the representation of spatial fea-tures at multiscale.Then we densely connect series of MSEFE or MSAFE modules respectively in a two-branch framework to balance efficiency and effectiveness,alleviate the vanishing-gradient problem and strengthen the feature propagation.To evaluate the effectiveness of the proposed method,the experimental results were carried out on bench mark HsI datasets,demonstrating that TBMSSN obtained higher classification accuracy compared with several state-of-the-art methods.展开更多
针对步态识别易受拍摄视角、外观变化等影响的问题,提出一种基于双支路卷积网络的步态识别方法。首先,提出随机裁剪随机遮挡的数据增强方法RRDA(Restricted Random Data Augmentation),以扩展外观变化的数据样本,提高模型遮挡的鲁棒性;...针对步态识别易受拍摄视角、外观变化等影响的问题,提出一种基于双支路卷积网络的步态识别方法。首先,提出随机裁剪随机遮挡的数据增强方法RRDA(Restricted Random Data Augmentation),以扩展外观变化的数据样本,提高模型遮挡的鲁棒性;其次,采用结合注意力机制的两路复合卷积层(C-Conv)提取步态特征,一个分支通过水平金字塔映射(HPM)提取行人外观全局和最具辨识度的信息;另一分支通过多个并行的微动作捕捉模块(MCM)提取短时间的步态时空信息;最后,将两个分支的特征信息相加融合,再通过全连接层实现步态识别。基于平衡样本特征的区分能力和模型的收敛性构造联合损失函数,以加速模型的收敛。在CASIA-B步态数据集上进行实验,所提方法在3种行走状态下的平均识别率分别达到97.40%、93.67%和81.19%,均高于GaitSet方法、CapsNet方法、双流步态方法和GaitPart方法;在正常行走状态下比GaitSet方法的识别准确率提升了1.30个百分点,在携带背包状态下提升了2.87个百分点,在穿着外套状态下提升了10.89个百分点。实验结果表明,所提方法是可行、有效的。展开更多
针对红外图像与可见光图像融合中细节丢失严重,红外图像的特征信息未能突出显示以及源图像的语义信息被忽视的问题,提出一种基于二次图像分解的红外图像与可见光图像融合网络(Secondary Image Decomposition For Infrared And Visible I...针对红外图像与可见光图像融合中细节丢失严重,红外图像的特征信息未能突出显示以及源图像的语义信息被忽视的问题,提出一种基于二次图像分解的红外图像与可见光图像融合网络(Secondary Image Decomposition For Infrared And Visible Image Fusion,SIDFuse)。利用编码器对源图像进行二次分解以提取不同尺度的特征信息,然后利用双元素注意力为不同尺度的特征信息分配权重、引入全局语义支路,再采用像素相加法作为融合策略,最后通过解码器重建融合图像。实验选择FLIR数据集用于训练,采用TNO和RoadScene两个数据集进行测试,并选取八种图像融合客观评价参数进行实验对比分析。由TNO数据集的图像融合实验表明,在信息熵、标准差、空间频率、视觉保真度、平均梯度、差异相关系数、多层级结构相似性、梯度融合性能评价指标上,SIDFuse比基于卷积网络中经典融合算法DenseFuse分别平均提高12.2%,9.0%,90.2%,13.9%,85.1%,16.8%,6.7%,30.7%,比最新的融合网络LRRNet分别平均提高2.5%,5.6%,31.5%,5.4%,25.2%,17.9%,7.5%,20.7%。可见本文所提算法融合的图像对比度较高,可以同时更有效保留可见光图像的细节纹理和红外图像的特征信息,在同类方法中占有明显优势。展开更多
基金supported by the NationalNatural Science Foundation of China(41402254)Department of Science and Technology of Shaanxi Province(2019ZDLSF07-0701)。
文摘Two branches of Tangjiagou rock avalanche were triggered by Lushan earthquake in Sichuan Province,China on April 20th,2013.The rock avalanche has transported about 1500000 m3 of sandstone from the source area.Based on discrete element modeling,this study simulates the deformation,failure and movement process of the rock avalanche.Under seismic loading,the mechanism and process of deformation,failure,and runout of the two branches are similar.In detail,the stress concentration occur firstly on the top of the mountain ridge,and accordingly,the tensile deformation appears.With the increase of seismic loading,the strain concentration zone extends in the forward and backward directions along the slipping surface,forming a locking segment.As a result,the slipping surface penetrates and the slide mass begin to slide down with high speed.Finally,the avalanche accumulates in the downstream and forms a small barrier lake.Modeling shows that a number of rocks on the surface exhibit patterns of horizontal throwing and vertical jumping under strong ground shaking.We suggest that the movement of the rock avalanche is a complicated process with multiple stages,including formation of the two branches,high-speed sliding,transformation into debris flows,further movement and collision,accumulation,and the final steady state.Topographic amplification effects are also revealed based on acceleration and velocity of special monitoring points.The horizontal and vertical runout distances of the surface materials are much greater than those of the internal materials.Besides,the sliding duration is also longer than that of the internal rock mass.
基金supported by the National Natural Science Foundation of China(62077038,61672405,62176196 and 62271374)。
文摘In the field of hyperspectral image(HSI)classification in remote sensing,the combination of spectral and spatial features has gained considerable attention.In addition,the multiscale feature extraction approach is very effective at improving the classification accuracy for HSIs,capable of capturing a large amount of intrinsic information.However,some existing methods for extracting spectral and spatial features can only generate low-level features and consider limited scales,leading to low classification results,and dense-connection based methods enhance the feature propagation at the cost of high model complexity.This paper presents a two-branch multiscale spectral-spatial feature extraction network(TBMSSN)for HSI classification.We design the mul-tiscale spectral feature extraction(MSEFE)and multiscale spatial feature extraction(MSAFE)modules to improve the feature representation,and a spatial attention mechanism is applied in the MSAFE module to reduce redundant information and enhance the representation of spatial fea-tures at multiscale.Then we densely connect series of MSEFE or MSAFE modules respectively in a two-branch framework to balance efficiency and effectiveness,alleviate the vanishing-gradient problem and strengthen the feature propagation.To evaluate the effectiveness of the proposed method,the experimental results were carried out on bench mark HsI datasets,demonstrating that TBMSSN obtained higher classification accuracy compared with several state-of-the-art methods.
文摘针对步态识别易受拍摄视角、外观变化等影响的问题,提出一种基于双支路卷积网络的步态识别方法。首先,提出随机裁剪随机遮挡的数据增强方法RRDA(Restricted Random Data Augmentation),以扩展外观变化的数据样本,提高模型遮挡的鲁棒性;其次,采用结合注意力机制的两路复合卷积层(C-Conv)提取步态特征,一个分支通过水平金字塔映射(HPM)提取行人外观全局和最具辨识度的信息;另一分支通过多个并行的微动作捕捉模块(MCM)提取短时间的步态时空信息;最后,将两个分支的特征信息相加融合,再通过全连接层实现步态识别。基于平衡样本特征的区分能力和模型的收敛性构造联合损失函数,以加速模型的收敛。在CASIA-B步态数据集上进行实验,所提方法在3种行走状态下的平均识别率分别达到97.40%、93.67%和81.19%,均高于GaitSet方法、CapsNet方法、双流步态方法和GaitPart方法;在正常行走状态下比GaitSet方法的识别准确率提升了1.30个百分点,在携带背包状态下提升了2.87个百分点,在穿着外套状态下提升了10.89个百分点。实验结果表明,所提方法是可行、有效的。
文摘针对红外图像与可见光图像融合中细节丢失严重,红外图像的特征信息未能突出显示以及源图像的语义信息被忽视的问题,提出一种基于二次图像分解的红外图像与可见光图像融合网络(Secondary Image Decomposition For Infrared And Visible Image Fusion,SIDFuse)。利用编码器对源图像进行二次分解以提取不同尺度的特征信息,然后利用双元素注意力为不同尺度的特征信息分配权重、引入全局语义支路,再采用像素相加法作为融合策略,最后通过解码器重建融合图像。实验选择FLIR数据集用于训练,采用TNO和RoadScene两个数据集进行测试,并选取八种图像融合客观评价参数进行实验对比分析。由TNO数据集的图像融合实验表明,在信息熵、标准差、空间频率、视觉保真度、平均梯度、差异相关系数、多层级结构相似性、梯度融合性能评价指标上,SIDFuse比基于卷积网络中经典融合算法DenseFuse分别平均提高12.2%,9.0%,90.2%,13.9%,85.1%,16.8%,6.7%,30.7%,比最新的融合网络LRRNet分别平均提高2.5%,5.6%,31.5%,5.4%,25.2%,17.9%,7.5%,20.7%。可见本文所提算法融合的图像对比度较高,可以同时更有效保留可见光图像的细节纹理和红外图像的特征信息,在同类方法中占有明显优势。
